alibaba/EasyCV

Health v2 · maintenance only

Adoption not part of the score

1954 stars · 227 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

EasyCV is an all-in-one PyTorch-based computer vision toolkit from Alibaba covering self-supervised learning, vision transformers, and major CV tasks like classification, object detection, metric learning, and pose estimation. It includes SOTA SSL algorithms (SimCLR, MoCo, Swav, DINO, MAE), pretrained transformer models, efficient multi-GPU training with DALI acceleration, and deployment support on PAI-EAS.

Use cases

• train self-supervised vision models with DINO or MAE
• run image classification with pretrained vision transformers
• fine-tune object detection models like DETR
• benchmark SSL model representations
• deploy CV models as online inference services
• accelerate data preprocessing for vision training

When to choose

• you need a unified PyTorch toolbox for SSL and transformer-based vision tasks
• you want off-the-shelf SOTA vision transformer models and pretrained weights
• you need efficient multi-GPU training with accelerated data loading
• you plan to deploy vision models as scalable online services

When to avoid

• you need tasks outside its supported areas like segmentation or video
• you prefer lightweight single-purpose libraries
• you work outside the PyTorch ecosystem
• you need heavy customization of the training framework

Facets

library · maturity active

machine-learning deep-learning image-processing computer-vision benchmarking computer-vision machine-learning deep-learning image-processing python self-supervised-learning vision-transformer object-detection image-classification pytorch metric-learning pose-estimation model-deployment linux gpu docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="alibaba/EasyCV")